Preview (finally) of datatable-autocomplete

[Michael O'Cleirigh <mi...@rivulet.ca>]

Hello,

I got my wicket stuff commit access over a year ago and finally have the 
datatable-autocomplete module ready to be committed into wicketstuff-core.

I have a working version that I will be committing as soon as I can 
confirm that wicketstuff-core will still build with my new module 
(Hopefully tomorrow or next week at the latest).

The main use for this extension is to have quick ajax lookups for large 
static datasets.   It includes a Patrica Trie 
(http://en.wikipedia.org/wiki/Radix_tree) which is the index and the 
necessary wicket components and behaviors to tie a text field into an 
IDataProvider to display the results in a data table.

I've deployed the example application on my website here:
http://rivulet.ca:8080/datatable-autocomplete-examples-1.4-SNAPSHOT/

This is the best test case I could come up with to generate a large 
number of items to search, in this case 74271.  It works by opening up 
the rt.jar file in the jvm classpath and then indexing all of the 
methods found.  You can search by the method name only but sort the 
results by method name, parameters or class name and see how many 
results there are in total after each character of the search.

There are two search modes:
1. prefix matching which is what the Trie is meant for and is the 
fastest at.
2. any string matching which is still fast at 100,000 elements but 
probably slows down faster than prefix matching as the dataset size or 
maximum indexed string length grows large.

Regards,

Mike

Re: Preview (finally) of datatable-autocomplete

[Igor Vaynberg <ig...@gmail.com>]

that looks pretty impressive...

-igor

On Thu, Dec 10, 2009 at 7:39 PM, Michael O'Cleirigh
<mi...@rivulet.ca> wrote:
> Hello,
>
> I got my wicket stuff commit access over a year ago and finally have the
> datatable-autocomplete module ready to be committed into wicketstuff-core.
>
> I have a working version that I will be committing as soon as I can confirm
> that wicketstuff-core will still build with my new module (Hopefully
> tomorrow or next week at the latest).
>
> The main use for this extension is to have quick ajax lookups for large
> static datasets.   It includes a Patrica Trie
> (http://en.wikipedia.org/wiki/Radix_tree) which is the index and the
> necessary wicket components and behaviors to tie a text field into an
> IDataProvider to display the results in a data table.
>
> I've deployed the example application on my website here:
> http://rivulet.ca:8080/datatable-autocomplete-examples-1.4-SNAPSHOT/
>
> This is the best test case I could come up with to generate a large number
> of items to search, in this case 74271.  It works by opening up the rt.jar
> file in the jvm classpath and then indexing all of the methods found.  You
> can search by the method name only but sort the results by method name,
> parameters or class name and see how many results there are in total after
> each character of the search.
>
> There are two search modes:
> 1. prefix matching which is what the Trie is meant for and is the fastest
> at.
> 2. any string matching which is still fast at 100,000 elements but probably
> slows down faster than prefix matching as the dataset size or maximum
> indexed string length grows large.
>
> Regards,
>
> Mike
>
>
>